Authors: Shreyas Seshadri, Tuomo Raitio, Dan Castellani, Jiangchuan Li
Abstract: The semantic information conveyed by a speech signal is strongly influenced by local variations in prosody. Recent parallel neural text-to-speech (TTS) synthesis methods are able to generate speech with high fidelity while maintaining high performance. However, these systems often lack simple control over the output prosody, thus restricting the semantic information conveyable for a given text. This paper proposes a hierarchical parallel neural TTS system for prosodic emphasis control by learning a latent space that directly corresponds to a change in emphasis. Three candidate features for the latent space are compared: 1) Variance of pitch and duration within words in a sentence, 2) a wavelet based feature computed from pitch, energy, and duration and 3) a learned combination of the above features. Objective measures reveal that the proposed methods are able to achieve a wide range of emphasis modification, and subjective evaluations on the degree of emphasis and the overall quality indicate that they show promise for real-world applications.
The following samples demonstrate the emphasis control capability of the VarianceBased, WaveletBased and Combined methods. Emphasis modification is done by increasing the predicted emphasis features by 0.75 for the WaveletBased and Combined methods and by 0.5 for both features of the VarianceBased method.
Text (emphasized word in bold)
Voice 1
Voice 2
VarianceBased
WaveletBased
Combined
VarianceBased
WaveletBased
Combined
The total average height of the troposphere is thirteen kilometers.
The total average height of the troposphere is thirteen kilometers.
The total average height of the troposphere is thirteen kilometers.
The total average height of the troposphere is thirteen kilometers.
The total average height of the troposphere is thirteen kilometers.
The total average height of the troposphere is thirteen kilometers.
The total average height of the troposphere is thirteen kilometers.